In this paper, we survey loop parallelization algorithms, analyzing th
e dependence representations they use, the loop transformations they g
enerate, the code generation schemes they require, and their ability t
o incorporate various optimizing criteria such as maximal parallelism
detection, permutable loop detection, minimization of synchronizations
, easiness of code generation, etc. We complete the discussion by pres
enting new results related to code generation and loop fusion for a pa
rticular class of multidimensional schedules called shifted linear sch
edules. We demonstrate that algorithms based on such schedules lead to
simple codes. (C) 1998 Elsevier Science B.V. All rights resented.